典型文献
Two-Stream Architecture as a Defense against Adversarial Example
文献摘要：
The performance of deep learning on many tasks has been impressive. However, recent studies have shown that deep learning systems are vulnerable to small specifically crafted perturbations imperceptible to humans. Images with such perturbations are called adversarial examples. They have been proven to be an indisputable threat to deep neural networks (DNNs) based applications, but DNNs have yet to be fully elucidated, consequently preventing the development of efficient defenses against adversarial examples. This study proposes a two-stream architecture to protect convolutional neural networks (CNNs) from attacks by adversarial examples. Our model applies the idea of "two-stream" used in the security field. Thus, it successfully defends different kinds of attack methods because of differences in "high-resolution" and "low-resolution" networks in feature extraction. This study experimentally demonstrates that our two-stream architecture is difficult to be defeated with state-of-the-art attacks. Our two-stream architecture is also robust to adversarial examples built by currently known attacking algorithms.
文献关键词：
中图分类号：
作者姓名：
Hao Ge;Xiao-Guang Tu;Mei Xie;Zheng Ma
作者机构：
School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu 610054;School of Electronic Science and Engineering, University of Electronic Science and Technology of China,Chengdu 610054
文献出处：
引用格式：
[1]Hao Ge;Xiao-Guang Tu;Mei Xie;Zheng Ma-.Two-Stream Architecture as a Defense against Adversarial Example)[J].电子科技学刊,2022(01):81-91
A类：
defeated
B类：
Two,Stream,Architecture,Defense,against,Adversarial,Example,performance,deep,learning,many,tasks,has,been,impressive,However,recent,studies,have,shown,that,systems,are,vulnerable,small,specifically,crafted,perturbations,imperceptible,humans,Images,such,called,adversarial,examples,They,proven,indisputable,threat,neural,networks,DNNs,applications,but,yet,elucidated,consequently,preventing,development,efficient,defenses,This,study,proposes,stream,architecture,protect,convolutional,CNNs,from,attacks,by,Our,model,applies,idea,used,security,field,Thus,successfully,defends,different,kinds,methods,because,differences,high,resolution,low,feature,extraction,experimentally,demonstrates,our,difficult,state,art,also,robust,built,currently,known,attacking,algorithms
AB值：
0.636595
相似文献
机标中图分类号，由域田数据科技根据网络公开资料自动分析生成，仅供学习研究参考。
